Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Links to blind listening tests: discussion (Read 27473 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Links to blind listening tests: discussion

Wow, impressive list. Thanks

Maybe this is a usefull addtion: http://seanolive.blogspot.com/
TheWellTemperedComputer.com

Links to blind listening tests: discussion

Reply #1
By Roberto J. Amorim. Old tests : some codecs have been improved since then.
http://www.rjamorim.com/test/index.html

By Sebastian Mares, who managed the public tests after Roberto.
http://www.listening-tests.info/

These sites are now gone. I have created a mirror of them: http://listeningtests.t35.com

Sebastian's site is also archived at HA: http://listening-tests.hydrogenaudio.org/sebastian/
Here is Sebastian's thread about closing his site: http://www.hydrogenaudio.org/forums/index....showtopic=76672

Links to blind listening tests: discussion

Reply #2
Thanks for the new link, Alex B. I corrected it.

Roseval, Sean's Blog is not a blind test in itself. But another section could be created with links of this kind. It could include

Sean Olive's Audio musings :
http://seanolive.blogspot.com/

Bruce Coppola's audio page
http://bruce.coppola.name/audio/wisdom.html

The Audio Critic :
http://www.theaudiocritic.com/

Boston Audio Society :
http://www.bostonaudiosociety.org/articles.htm

David Griesinger's website :
http://www.davidgriesinger.com/

etc.

What do you think ?

Links to blind listening tests: discussion

Reply #3
I would also like to say that any moderator or administrator is welcome to update the pinned post with new links.

The update should be notified either as an answer to the pinned post, or here, so that we don't add the same links twice.

A line stating "last updated by xxxx on dd/mm/yyyy" should also be added on top of the post, and be updated by anyone posting modifications in the list.

Links to blind listening tests: discussion

Reply #4
I would also like to say that any moderator or administrator is welcome to update the pinned post with new links.

The update should be notified either as an answer to the pinned post, or here, so that we don't add the same links twice.

A line stating "last updated by xxxx on dd/mm/yyyy" should also be added on top of the post, and be updated by anyone posting modifications in the list.

Sorry if I'm just not seeing it, but I can't see a link here to the actual pinned topic referred to. I assume it's this: http://www.hydrogenaudio.org/forums/index....showtopic=82777

Links to blind listening tests: discussion

Reply #5
Yes, that's it.

Links to blind listening tests: discussion

Reply #6
I am very happy to have recently discovered HA. Especially after some long discussions with friends of mine who insisted that they can hear a difference between FLAC and a high quality MP3, but were not willing to perform a simple blind test. It's a pleasure to see that so much effort is put here into objective blind listening tests.

However, I am surprised to find so little academic research on in this subject. The problem with most ABX tests on this forum is that the number of participants is generally limited and that you are dependent on the equipment of the participants.  I find it strange that I can't find any academic research into this subject. I would love to see a test involving a large number of participants, a fixed set of (high quality) audio equipment and proper statistical analysis. With a  large number of participants also factors like age (with their differing audible frequency range) could be investigated.

Considering that many codecs originate from academic institutes (i.e. Fraunhofer etc.) I expected that there would be a large number of scientific experiments on the performance of different codecs. However, I cannot find any of them. Am I perhaps looking in the wrong direction?

 

Links to blind listening tests: discussion

Reply #7
Thanks for a useful bump. The above mentioned listeningtests.t35.com link is dead.  The correct address is http://listening-tests.freetzi.com :

FYI,

My mirror site is now available only at http://listening-tests.freetzi.com . The t35.com hosting service ended a few days ago.


Pio2001/other moderators, could you please fix the link in this post: http://www.hydrogenaudio.org/forums/index....mp;#entry717387 . Thanks.

Links to blind listening tests: discussion

Reply #8
@dirtyharry
I guess the chief point is that academic research, especially research involving large numbers of subjects, needs funding, and there just isn't anything to be found out that would tempt a funding body to open its money-chest. It's unlikely that anyone is going to make new and exciting findings.

Reputable manufacturers do undertake blind testing to improve their products, especially speakers--Sean Olive's blog is good on that, and some of the findings are published there, but naturally they're not going to make a gift of their research to their competitors (that is to say, such of the rival manufacturers who try to sell on SQ, as opposed to woo and waving of the dollar-willy).

tl;dr: there's not much money in the academic world, and the competition for funding is horrendous.

Links to blind listening tests: discussion

Reply #9
I am very happy to have recently discovered HA. Especially after some long discussions with friends of mine who insisted that they can hear a difference between FLAC and a high quality MP3, but were not willing to perform a simple blind test. It's a pleasure to see that so much effort is put here into objective blind listening tests.

We have all been there.
Quote
However, I am surprised to find so little academic research on in this subject. The problem with most ABX tests on this forum is that the number of participants is generally limited and that you are dependent on the equipment of the participants.  I find it strange that I can't find any academic research into this subject. I would love to see a test involving a large number of participants, a fixed set of (high quality) audio equipment and proper statistical analysis. With a  large number of participants also factors like age (with their differing audible frequency range) could be investigated.

I think that the tests carried out by Floyd Toole in the 80s are good examples of how it can be done (loudspeakers in small rooms). The ideal test depends on what it is that you want to know. Do you want to figure out what the general public is able to distinguish? Do you want to estimate an upper bound to what 1/1000 or 1/100000 of the population are able to distinguish? Or do you want to know what you yourself is able to distinguish?

-k

Links to blind listening tests: discussion

Reply #10
Hi !

I've added two ABX tests to the pinned topic.

An impressive result, although some successes of the same kind existed before :

DA-AD chain (Xonar Essence STX, Xonar D1 soundcards), with different antialias filter settings. Software emulation of the soundcard filter included. Successful ABX for all files except the dummy one.

Results : http://www.head-fi.org/t/676885/successful...ed-to-192-24/60
Description of the files : http://www.head-fi.org/t/676885/successful...90#post_9738961


A much less impressive test :

Turntables :

[French] Luxman PD-282 vs Technics SL-3100 (both recorded digitally). Failure
http://www.homecinema-fr.com/forum/materie...-t30037817.html

Links to blind listening tests: discussion

Reply #11
Hi !

I've added two ABX tests to the pinned topic.

An impressive result, although some successes of the same kind existed before :

DA-AD chain (Xonar Essence STX, Xonar D1 soundcards), with different antialias filter settings. Software emulation of the soundcard filter included. Successful ABX for all files except the dummy one.

Results : http://www.head-fi.org/t/676885/successful...ed-to-192-24/60
Description of the files : http://www.head-fi.org/t/676885/successful...90#post_9738961


If I understand this test correctly, it is a comparison between a file and an upsampled version of itself. If that is true, do everything right and there can be no audible difference. Sort of like comparing a flac file to the wav file it was made from.

I would say that it is an enigmatic test, because if you do it right with no failures of your experiment or the equipment in your experiment, you should obtain random guessing. If you obtain a positive result, then it proves that there is a failure in your experiment.

Links to blind listening tests: discussion

Reply #12
Mr Atkinson would exclaim a "newfound sense of spaciousness and instrument seperation" due to said up sampling. Yes I have a sub for stereophile...its 0.99 per month and worth every penny! The "National Inquirer" of the audio world...


Links to blind listening tests: discussion

Reply #14
If I understand this test correctly, it is a comparison between a file and an upsampled version of itself.


No. The test is between a reference file and five other. They are all 24 bits 192 kHz, and they are all created from a 16 bits 44.1 kHz one.

Two files are properly upsampled, the reference and the dummy one. They are identical.
Three ones are analog copies of the 16/44.1 one. One from a Xonar Essence STX soundcard, and two from a Xonar D1 soundcard used with two different antialias filters during the D/A conversion.
One is a digital simulation of the Xonar D1 slow roll-off D/A filter.

Though the double blind trials were run rather chaotically, a description of the method by the tester seems to show that the results are sound.

if you do it right with no failures of your experiment or the equipment in your experiment, you should obtain random guessing.


So it seems, but here, the goal was not to do it right. It was to test the result when you do it with standard equipment.

Strangely enough, while the insertion of an AD/DA step in the path of an analog signal is nearly never detected in blind listening tests, it is not uncommon to see the application of a DA/AD step on a digital file being ABXed with success.

Links to blind listening tests: discussion

Reply #15
Strangely enough, while the insertion of an AD/DA step in the path of an analog signal is nearly never detected in blind listening tests, it is not uncommon to see the application of a DA/AD step on a digital file being ABXed with success.


Nothing strange about that to me. No analog source other than a direct microphone or console feed from an actual live performance has the low distortion, flat frequency response, and excellent noise performance of digital media. Not analog tape and certainly not LP.

For a DA/AD step to be ABXed successfully these days takes very substandard converters, because even mediocre converters are known to pass.

Try the Sound Blaster samples here: http://ethanwiner.com/aes/